DigestAI news desk

Cut through the AI noise.

Hardware & Compute2 min read

Anthropic appears to block Huawei's 950DT and Amazon's Tranium3 chips with Opus 5.5

Anthropic’s Claude Opus 5.5 model includes classifiers that can cause the system to fall back to Opus 5. According to the model’s support document, these classifiers limit frontier‑LLM kernel development on a small set of accelerators. An X user says the classifiers specifically target Huawei’s 950DT AI chip and Amazon’s Tranium3 custom AI chip. Anthropic says the restrictions should not affect…

1 source

Key points

  • Claude Opus 5.5’s support doc lists classifiers that block Huawei’s 950DT chip and Amazon’s Tranium3 chip for frontier LLM kernel work
  • Classifiers cause Opus 5.5 to fall back to Opus 5 and also prevent full reasoning extraction
  • Anthropic has not clarified if the Amazon chip restriction is deliberate; the reason for Huawei’s block is clearer

The reason for the Huawei restriction is clear, but whether the Amazon block is intentional or accidental remains uncertain. The company has not commented further, and the impact on developers using those chips for cutting‑edge model development is still being assessed.

Model page: Claude Opus 5.5 →

The story so far

5 episodes →
  1. Anthropic appears to block Huawei's 950DT and Amazon's Tranium3 chips with Opus 5.5this story
Full story from wccftech.com · by Ramish Zafar · via Search: AnthropicOpen source ↗

Anthropic Blocks Huawei Chips From Using Latest Opus 5.5 to Develop AI Models, Yet Amazon Appears To Have Gotten Caught in the Crossfire

wccftech.com · 23 September 2026

Loading the full article…

This text was published by wccftech.com and written by Ramish Zafar. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Topics · follow one to build your own front page

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Hardware & Compute

All →

Related stories